With the development and popularization of a virtual reality (VR) panoramic video technology, a visual angle of a picture is changed by means of head tracking, so that a visual system and a motion perception system of a user are connected, and the user can observe a presented scene through head motion. A VR panoramic video device enables the user to watch a video at any visual angle within 360 degrees. In order to bring stereo perception experience to users by using the VR panoramic video technology, an audio recording technology in a panoramic video also becomes a focus, and how to record audio that can be synchronous with an image becomes an important research topic.
In the VR panoramic video technology, in order to achieve a verisimilar hearing effect, in a video playing process, an image in each direction requires audio that is recorded in the corresponding direction. If there is no corresponding audio recording mechanism, when a user faces a certain direction, heard sounds are possibly sounds recorded when the user faces another direction, and stereo experience of synchronization between an image and a sound in the VR panoramic video technology cannot be truly experienced. However, an existing audio recording device usually performs recoding at a point by using an omni-directional microphone, a sound is single in source and has flat and non-stereoscopic tone quality, or only audio in four directions: front, back, left, and right are picked up. During watch of a VR panoramic video, an average transition angle for turning from a sound field in one direction to a sound field in another direction in a human head rotation process is 90 degrees. Therefore, a span is relatively large, that is, a fusion band is relatively large, a match effect with an image is poor, and a requirement for synchronization between a sound and an image in the VR panoramic video technology cannot be met.